Type of Document	Dissertation
Author	Marinovici, Laurentiu Dan
URN	etd-05202011-111953
Title	Communication-constrained Feedback Stability and Multi-agent System Consensusability in Networked Control Systems
Degree	Doctor of Philosophy (Ph.D.)
Department	Electrical & Computer Engineering
Advisory Committee	Advisor Name Title Zhou, Kemin Committee Chair Gu, Guoxiang Committee Co-Chair Wei, Shuangqing Committee Member Wolenski, Peter Committee Member Sathivel, Subramanian Dean's Representative
Keywords	communication constrained feedback multi-agent systems consensusability logarithmic quantizer networked control systems
Date of Defense	2011-05-12
Availability	unrestricted
Abstract With the advances in wireless communication, the topic of Networked Control Systems (NCSs) has become an interesting research subject. Moreover, the advantages they offer convinced companies to implement and use data networks for remote industrial control and process automation. Data networks prove to be very efficient for controlling distributed systems, which would otherwise require complex wiring connections on large or inaccessible areas. In addition, they are easier to maintain and more cost efficient. Unfortunately, stability and performance control is always going to be affected by network and communication issues, such as band-limited channels, quantization errors, sampling, delays, packet dropouts or system architecture. The first part of this research aims to study the effects of both input and output quantization on an NCS. Both input and output quantization errors are going to be modeled as sector bounded multiplicative uncertainties, the main goal being the minimization of the quantization density, while maintaining feedback stability. Modeling quantization errors as uncertainties allows for robust optimal control strategies to be applied in order to study the accepted uncertainty levels, which are directly related to the quantization levels. A new feedback law is proposed that will improve closed-loop system stability by increasing the upper bound of allowed uncertainty, and thus allowing the use of a coarser quantizer. Another aspect of NCS deals with coordination of the independent agents within a Multi-agent System (MAS). This research addresses the consensus problem for a set of discrete-time agents communicating through a network with directed information flow. It examines the combined effect of agent dynamics and network topology on agents' consensusability. Given a particular consensus protocol, a sufficient condition is given for agents to be consensusable. This condition requires the eigenvalues of the digraph modeling the network topology to be outer bounded by a fan-shaped area determined by the Mahler measure of the agents' dynamics matrix.
Files	Filename       Size       Approximate Download Time (Hours:Minutes:Seconds)     28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access    marinovicidiss.pdf 1.38 Mb 00:06:22 00:03:16 00:02:52 00:01:26 00:00:07

If you have questions or technical problems, please Contact LSU-ETD Support.